Infinite product decomposition of orbifold mapping spaces (Q1017842): Difference between revisions

Let \(G\) be a finite group and \(M\) be a \(G\)-manifold. By the orbifold loop space is meant the space \(L_{\text{orb}}(M/G)\) given by \(L_{\text{orb}}(M/G)=(\amalg_{g \in G}L_gM)/G\), where \(L_gM=\{ \gamma : \mathbb{R} \to M \;| \;\gamma(t+1) = g^{-1}\gamma(t) \;\text{for \;all} \;t \in \mathbb{R} \}\). Then the orbifold elliptic genus \(\text{ell}_{\text{orb}}(M/G)\) of \((M, G)\) is defined as the \(S^1\)-equivariant Hirzebruch \(\chi_y\)-characteristic of \(L_{\text{orb}}(M/G)\), i.e. \(\text{ell}_{\text{orb}}(M/G)=\chi_y^{S^1}(L_{\text{orb}}(M/G)) \in \mathbb{Z}[y, y^{-1}][[q]]\), where the powers of \(q\) are characters of \(S^1\), and then further one knows that there holds the formula \[ \sum_{n \geq 0}p^n\text{ell}_{\text{orb}} (\text{SP}^n(M/G)) = \prod_{n \geq 1, m \geq 0, k \in \mathbb{Z}}(1 - p^nq^my^k)^{-c(mn, k)} \] where \(\text{ell}_{\text{orb}}(M/G)=\sum_{m \geq 0, k \in \mathbb{Z}} c(m, k)q^my^k\) [\textit{R. Dijkgraaf, G. Moore, E. Verlinde} and \textit{H. Verlinde}, Commun. Math. Phys. 185, No. 1, 197--209 (1997; Zbl 0872.32006); \textit{L. Borisov} and \textit{A. Libgober}, Duke Math. J. 116, No. 2, 319--351 (2003; Zbl 1053.14050)]. Here \(\text{SP}^n(X)\) denotes the \(n\)-th symmetric product \(X^n/{\mathfrak S}_n\) of a space \(X\). This paper seems to be motivated by an attempt to present a geometric interpretation of this formula. For a variable \(t\), let \(S_t(X)=\Sigma_{k\geq 0}t^k\text{SP}^k(X)\) where the summation symbol means topological disjoint union. The author shows that actually one has \[ \sum_{n\geq 0}p^nL_{\text{ord}}(\text{SP}^n(M/G)) \cong \prod_{r\geq 1}S_{p^r}(L_{\text{ord}}^{(r)}(M/G)/\mathbb{Z}_r) \] where \(L_{\text{ord}}^{(r)}(M/G)\) is the space of orbifold loops of length \(r\). This is just a required geometric realization of the above formula. In this paper, however, this geometric decomposition formula is considered in a more general context in which a certain mapping space \(\text{Map}_{\text{orb}}(\Sigma, M/G)\) is used instead of an orbifold loop space (Theorem A). Here \(\Sigma\) denotes an arbitrary connected manifold and then the above geometric decomposition formula corresponds to the case of \(\Sigma=S^1\). It is remarked that this generalized formula can be used to recover the author's previous results in [Algebr. Geom. Topol. 1, 115--141 (2001; Zbl 0965.57033) and ibid. 3, 791--856 (2003; Zbl 1037.57022)], and in fact some of them are reproduced here. Furthermore, it is shown that this formula leads one to the idea of geometric Hecke operators for functors, which satisfy the Hecke identity for 2-dimensional tori (Theorem B).